Can the tumor genome help us to better select

patients for antiangiogenic therapy ?

Emile E Voest

Department of Medical Oncology

UMC Utrecht Cancer Center

TAT meeting Paris 2011

Ocaña A, et al. J Clin Oncol. 2011;29:254-6

Modest succes of bevacizumab

in common cancers

Can we afford to develop drugs the way we

used to ?

How Much Is Life Worth: Cetuximab,

Non–Small Cell Lung Cancer, and the

$440 Billion Question

Tito Fojo and Christine Grady

J Natl Cancer Inst 2009; 101:1044-1048

Discrepancy between preclinical models

and patients

• In vitro models commonly use cell lines that have been

maintained in culture for years

• Investigations focus on knock down of genes by siRNA,

but functional consequences of mutations are rarely

studied in preclinical models

• In vitro sensitivity to agents does not accurately predict

their activity in vivo

• Mouse models employ xenografts of selected cell lines

• Transgenic mouse models have a defined genetic initiator

of cancer

• Therefore: huge gap between preclinical models and our

patients!!

The patient as readout:

Personalized Cancer Treatment

Patient’s

tissue sample

pathology

grade, size, IHC

Chemotherapy

Tailored therapy

Patient’s

tissue sample

Targeted therapyPP

P

PP PPP

P P P P P P

PP

P

Molecular

diagnostics

Which pathways are active?

Proof of concept !

• After screening tumor samples from approximately 1500patients with non–small-cell lung cancer for the presence of

ALK rearrangement, 82 patients with advanced ALK-

positive disease were identified

• Mean treatment duration of 6.4 months

• Overall response rate : 57% (47 of 82 patients);

• 27 patients (33%) had stable disease.

• The estimated probability of 6-month progression-free

survival was 72%, with no median for the study reached.

Kwak EL, et al. N Engl J Med. 2010;363:1693-703.

Proof of concept !

EML4-ALK Mutations in Lung Cancer That Confer

Resistance to ALK Inhibitors

• Two secondary mutations within the kinase domain of

EML4-ALK in tumor cells isolated from a patient developed

during the relapse phase of treatment with an ALK inhibitor.

• Each mutation developed independently in subclones of

the tumor and conferred marked resistance to two different

ALK inhibitors.

Choi YL. et al. N Engl J Med 2010; 363:1734-1739

Can we achieve the same success for

anti-angiogenic therapy ?

Question that needs to be addressed:

How to determine genetic dependency

or “angiogene addiction” ?

“Angiogene addiction”

• Tumor acquires genetic changes in growth factor

pathways early in tumorigenesis that facilitate tumor

growth

• Short time between initial mutagenesis and clinical

manifestations ?

• Examples: clear cell renal cell cancer; pancreatic

NET, PEComa

kdr-like:egfp

Mutations in the VHL gene to select patients for

anti-angiogenic therapy

VHLElC

Cul2

E2

Rbx

ElB

Ub

HIF

UbUbUb

UbUb

UbUb

Ub

degradation

Rooijen et al. Blood 2010

Nordstrom-O’Brien et al., Human Mutations 2010

Rooijen et al., Dis Models Mech 2010

kdr-like:egfp

Mutations in the VHL gene to select patients for

anti-angiogenic therapy

sibling

vhl-/-

3D reconstruction

overlay

7.25 dpf, 25 hours post TAMRA injection

Rooijen et al. Blood 2010

Nordstrom-O’Brien et al., Human Mutations 2010

Rooijen et al., Dis Models Mech 2010

von Hippel-Lindau gene status and response to

VEGF targeted therapy for metastatic ccRCC

• 700 different types of mutations, derived from this protein of only

213 amino acids.

• Retrospective analysis in 123 ccRCC patients. Patients with VHL

inactivation had a response rate of 41% vs 31% for those with wild-

type VHL (p = 0.34). Patients with loss of function mutations had a

52% response rate vs 31% with wild-type VHL (p = 0.04).

• On multivariate analysis the presence of a loss of function mutation

remained an independent prognostic factor associated with

improved response.

• In VHL disease, clear evidence supports strong genotype-phenotype

correlations, but the situation in sporadic ccRCC is less clear.

Choueiri TK et al, J Urol 2008, Gossage L, Eisen T. Nat Rev Clin Oncol. 2010,

Nordstrom-O’Brien, et al. Human Mutations 2010

An other example of “angiogene addiction” ?

Comparative Genomic Hybridization of PEComa’s Pan et al, Human Pathology 2006

Green = chromosomal gain

Red = chromosomal loss

16p=TSC2 locus

Perivascular Epithelioid Cell tumor

Proof of concept…

• Clinical Activity of mTOR Inhibition With Sirolimus in Malignant Perivascular

Epithelioid Cell Tumors: Targeting the Pathogenic Activation of mTORC1 in

Tumors. Wagner et al. J Clin Oncol Feb 2010

PhosphoS6 staining

A clinical example of a successful DNA guided

anti-angiogenic treatment

• Patient with a perivascular epitheloid cell tumor

(PEComa) with liver and intraperitoneal metastases

• Initial response on doxorubicin but progressive

disease within 6 months, no other treatment options

January 2010 February 2011

“Angiogene addiction” in common cancer?

• Potential examples of “angiogene addiction” : ccRCC,

GIST, PEComa, pancreatic NET

• Better selection of patients with common cancers is

essential

• Can tumor DNA provide answers ?

Can we achieve the same

success in common cancers ?

Hanahan and Weinberg, Cell, april 2011

K-RAS

ACF

The Adenoma-Carcinoma Sequence in Colorectal Cancer:

tumor initiation, progression, and metastasis

intestinal

epithelial cell

p53

chr. 17p LOH

SMAD2/4

chr. 18q LOH

carcinoma adenoma

APC

metastasis

Genetic alterations

Are we sure that the target is present in metastatic disease ?

Primary tumors are not the same as metastases

• Paired colorectal tumors and metastasis of 21 patients

were selected (42 tumor samples)

• Tumor DNA was isolated by microdissection and

sequenced for 1263 genes covering all major pathways

and kinases

• Looking for function altering genetic changes

• Very stringent SNP calling

• On average a metastasis gains 89 relevant

mutations and loses 70 mutations present in

the primary

Conclusions

• Molecular diagnostics (sequencing, molecular imaging) will

improve patient selection for targeted therapy

• Single agent activity of anti-angiogenics is most likely seen in

tumors addicted to mutations in growth factor pathways

• Mutations, copy number variations, structural variance all

contribute

• Next generation sequencing data within clinical trials need to be

connected through data sharing (open source !!!)

• Biopsies of tumor at start of and at progression under treatment

are essential

Acknowledgements

- Marco Koudijs

- Sander Schouten

- Joost Vermaat

- Martijn Lolkema

- Wijnand Roessingh

- Nicolle Besselink

- Ilse van Oosterom

- Oscar Paling

- Laura Daenen

- Julia Houthuijzen

- Jeanine Roodhart

- Marlous Hoogstraat

- Stef van Lieshout

- Geert Cirkel

- Christa Hooijdonk

Amsterdam: Jan Schellens, Rene Bernards

Rotterdam: Stefan Sleijfer, John Martens

Utrecht: Edwin Cuppen, Emile Voest